CLASSIFICATION OF MALARIA INFECTED CELLS USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

Malaria is a life-threatening disorder resulting from parasites which might be transmitted to humans thru the bites of inflamed mosquitoes. Automation of the analysis method will allow correct analysis of the disease and subsequently holds the promise of turning in dependable health- care to resource-scarce areas. Machine mastering technologies were used for computerized prognosis of malaria. We present number of our latest progresses on relatively accurate type of malaria-inflamed cells with the use of deep convolutional neural networks. First, we describe image processing techniques used for segmentation of red blood cells from entire slide images. Secondly, the procedures of compiling a pathologists-curated image dataset for training deep neural network, as well as data augmentation methods used to noticeably enhance the scale of the dataset, in mild of the overfitting trouble related to training deep convolutional neural networks. Lastly, compare the classification accuracies obtained by deep convolutional neural networks through training, validating, and testing with various combinations of the datasets. These datasets consist of the authentic dataset and the notably augmented datasets, which might be acquired using direct interpolation, in addition to indirect interpolation with the use of routinely extracted features furnished through stacked autoencoders.